Skip to content

Registry-Mediated Discovery

Version
v2 · 2026-09-28 · History
Prime #
800
Domain group
Applied Sciences & Engineering
Origin domain
Computer Science & Software Engineering
Subdomain
Distributed Systems → Computer Science & Software Engineering

Core Idea

Agents find each other through a shared registry rather than by direct reference: providers register under stable logical names, consumers resolve those names to a current location at use-time, and the registry is the only address either side must know in advance — so the coupling survives turnover.

How would you explain it like I'm…

Ask the Front Desk

Say you want to visit your friend Mia at a big school, but you don't know which room she's in. You go to the front desk and ask, "Where is Mia?" and they tell you her room today. If Mia changes rooms tomorrow, the front desk just writes down the new room, and you can still find her by asking.

The Where-Is-Everyone List

Imagine a school where classes switch rooms all the time. Instead of every student memorizing where every teacher is today, everyone just goes to the school office and asks, "Where is Ms. Lopez's class?" Teachers tell the office whenever they move, so the office always has the current room. The teacher's name never changes, but the room it points to can, and nobody else has to update anything. The only place everyone needs to know by heart is the office.

Finding Things by Registry Lookup

Registry-Mediated Discovery is when parties find each other through a shared registry rather than by holding each other's addresses directly. It has three separate roles: providers, who register themselves under a stable name; consumers, who look up that name when they need it; and the registry, which maps each name to wherever the provider currently is. The name stays the same even when the location behind it changes, so a provider can move, leave, or be replaced without every consumer updating their own records. Instead of everyone tracking everyone (which grows fast as the group grows), everyone only needs to know where the registry is. This is different from just having names - a name alone doesn't tell you where something is now. A quick test: can a provider move without the people who depend on it having to change anything? If yes, discovery is going through a registry.

 

Registry-Mediated Discovery is a structural pattern of indirection: consumers locate providers by resolving a stable logical identifier against a shared registry, rather than by holding a direct reference to the provider. The structure is explicitly three-party - provider, consumer, and registry are distinct roles, and the registry is its own addressable entity, not a feature of either side. Providers register under a name; consumers hold only that name and resolve it at use-time to get the provider's current endpoint, location, or contract. The key property is a mutable mapping behind a stable name: the binding can change without consumers noticing. This replaces the N-by-N problem of every party tracking every other with a single point of indirection whose address is the only thing anyone must know in advance. As a result, the coupling survives turnover - providers can join, leave, and relocate without changes to consumer code or address books. DNS, service registries in microservice architectures, and a company switchboard are all instances. It should be distinguished from naming (merely having identifiers), from interfaces (how parties interact once connected), and from networking (the substrate that carries the lookup). The diagnostic question is whether a provider can change location without any consumer-side change; if not, the system relies on hardcoded references.

Broad Use

  • DNS: domain names resolve to current IP addresses; servers change IPs without client reconfiguration.
  • Microservices: services register endpoints under logical names; consumers query and receive an up-to-date endpoint.
  • Civic life: phone directories and library catalogs map stable names or call numbers to current numbers and shelf locations.
  • Scholarship: DOIs and bibliographic keys map to current canonical locations.
  • Platforms: drivers and hosts register availability; the matchmaking platform mediates discovery.
  • Microbiology: in quorum sensing, a chemical-concentration field acts as a passive registry of population density.

Clarity

Naming the pattern surfaces the registry as a separate object of design — who runs it, how it fails, who governs registration — instead of folding it into one of the parties it mediates.

Manages Complexity

It collapses N×M pairwise coupling to N+M+1: each side knows only the registry, so providers and consumers change independently through one indirection point.

Abstract Reasoning

When N parties must reach M, factor the stable-name/mutable-location mapping into a registry — and treat its three pathologies (bottleneck, trust-anchor, staleness) as the inherent price, to be bounded rather than wished away.

Knowledge Transfer

  • Distributed systems → healthcare: the registry refactor that rescues a tangled microservice config also rescues a referral network when personal address books break on staff turnover.
  • Engineering → any substrate: a practitioner who has managed DNS staleness asks, everywhere, what the staleness window is and who controls registration.
  • Architecture → microbiology: recognizing a passive concentration field as an implicit registry ports the coupling analysis to diffusion-mediated systems no one designed.

Example

A web operator renumbers from one IP to another and every browser worldwide reaches the new address without a single client change — the turnover-survival test that defines registry-mediated discovery.

Relationships to Other Abstractions

Local relationship map for Registry-Mediated DiscoveryParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Registry-MediatedDiscoveryPRIMEPrime abstraction: Indirection — is a kind ofIndirectionPRIMEDomain-specific abstraction: Deeds registration — is a kind ofDeedsregistrationDOMAIN

Current abstraction Registry-Mediated Discovery Prime

Parents (1) — outgoing typed relations

  • Registry-Mediated Discovery is a kind of Indirection Prime

    The specific three-party form of indirection — provider, consumer, and a registry that is its own addressable entity — whose payoff is turnover survival (a provider can relocate without consumer change).

Children (1) — abstractions linked here by typed relations

  • Deeds registration Domain-specific is a kind of Registry-Mediated Discovery

    It remains its own entry because its identity is fixed by the jurisdiction and date, parcel description, parties and estate, instrument type and execution, acknowledgment and taxes, registry and indexing method, recording time, chain of title, notice and priority statute, unrecorded interests, search and abstract, title insurance and transition to title registration.

Hierarchy paths (3) — routes to 3 parentless roots

Not to Be Confused With

  • Registry-Mediated Discovery is not Indirection because it is the specific three-party registry form whose payoff is turnover survival, whereas indirection is the general move of resolving a reference through a pointer.
  • Registry-Mediated Discovery is not Traceability because it is forward-looking resolution of a name to a current location, whereas traceability links artefacts backward to their origins for accountability.
  • Registry-Mediated Discovery is not Interface because it governs how parties find each other, whereas an interface governs how they interact once found.